Table of contents
Open Table of contents
Overview
- Client: Altum Inc. (Startup)
- Duration: Nov. 2019 – Oct. 2020
Summary
- Collaboratively created 6 crucial features to assist with faster article delivery.
Roles
- UX Designer, UI Designer, User Researcher
Background
With complaints about our articles’ quality and delivery time, Altum looked to research, design, and test AI tools for its content writing service, Jenni AI, that would help improve Jenni’s word processor.
A list of client complaints we compiled into our sprint report.
Goals
We hypothesized that proper AI features would help content writers deliver high quality articles faster and decrease the number of client complaints about late articles.
Process
The steps of Altum’s design process are:
- Define. Defining our users and the MVP design.
- Ideate. Ideating on design possibilities.
- Test. Testing pre-release ideations.
- Observe. Observing post-release outcomes.
The Altum design process.
Define
Our first move was to interview content writers who noted issues such as:
- Difficulties in adapting references into articles.
- Lowering the plagiarism score.
Excerpt of user interview notes regarding trouble with lowering the plagiarism score.
Ideate
I shared low and high-fidelity wireframes for feedback on each AI tool.
The AI tools addressed the various problems writers faced and addressed business needs as well.
The Paraphrasing tool that restructured existing text into new plagiarism-free text.
The Readability tool tackled the issue of lowering plagiarism scores by giving writers clear feedback.
The Simplifier also expedited lowering plagiarism scores by rewording input text to simplify it.
The Scraper automated referencing articles by copying text directly into the word processor.
The Express modal, while not an AI tool, organized the AI tools in a more intuitive way.
Test
We optimized for the most results with the least investment by conducting budget usability tests.
A screenshot of a user test with the subject completing tasks based on a scenario.
Observe
We tracked:
- The number of late articles to measure writing time.
- The number of client complaints.
Subsequently our results showed that:
- Overall writing time didn’t go down.
- The number of client complaints didn’t go down.
Reflection
The AI-Powered Word Processor didn’t meet its success metrics because the field of designing for AI usability is still new and the technology itself has a ways to go.
A sample of the many struggles Altum faced to improve Jenni’s AI tools.
Next steps
My next steps would’ve been to continue researching into improving the usability of our AI tools.